Abstract
This paper investigates the efficacy of S&P Dow Jones Indices’ (S&P DJI’s) glass-box optimization algorithm for incorporating multiple sustainability-related objectives in the construction of an index. The glass-box optimization underpins the S&P Paris-Aligned & Climate Transition, Sustainability Enhanced, ESG Enhanced and Net Zero Carbon Budget Indices. The approach is motivated by the growing demand for transparency in how sustainability, environmental, social and governance (ESG) and other climate-related objectives are incorporated into the index construction process, and an acknowledgement of the multi-faceted nature of these objectives. The glass-box optimization is compared to a representative risk model-based index optimization in three different scenarios, where each scenario is characterized by a different combination of constraints. Special emphasis is given to the interpretability and explainability of the optimized index weights, with the motivation to minimize the possibility of greenwashing that may be caused by insufficient association between the optimized index weights and the company characteristics used to define the constraints. The results provide strong support in favor of the glass-box optimization as a method for building sustainability indices. The index weights produced by the glass-box optimization are shown to be completely explainable in terms of the constraints, whereas the weights produced by the risk model optimization were strongly influenced by additional factors that are included in the model to explain the covariance matrix of returns. The indices derived by the glass-box optimization also relied less heavily on extreme positions in small, illiquid assets, while achieving similar levels of performance with respect to realized tracking error and lower portfolio turnover.
Introduction
The past decade has been marked by a significant increase in the proportion of wealth held in passive, index-based investment strategies, with 42.9% of assets held by U.S. mutual funds and ETFs now being managed passively, reflecting a 2.3% annualized increase since 2013. At the same time, the demand for sustainability and net zero emissions-aligned investment solutions has increased substantially. The growth in demand for ESG and climate-oriented investment strategies has been driven by the search for long-term financial value and the pursuit of investment opportunities that align with global sustainability objectives. This trend shows no sign of slowing, with PricewaterhouseCoopers (2022) reporting that 8 in 10 investors plan to boost their exposure to ESG over the next two years, and that by 2026, nearly USD 34 trillion in global assets will be directed into ESG funds and other sustainability investment vehicles.
The sustainability characteristics of an index can be improved by applying a series of stock-level selection criteria targeted at removing the least sustainable companies (e.g., business activity exclusions and best-in-class constituent selection) or by defining an index weighting scheme that allocates the greatest weight to companies with the most favorable characteristics (e.g., tilting strategies and constrained index optimization techniques). These steps can be applied individually or together as part of a multi-step methodology. This paper is principally concerned with the latter. Specifically, we consider the problem of improving the sustainability characteristics of a global stock market index by optimizing the weights subject to one or more constraints. This approach is consistent with recent research by Kölbel et al. (2020), who identified capital allocation, shareholder engagement and indirect impacts as the three channels through which sustainability investing contributes to societal goals. The effectiveness of the capital allocation mechanism relies on a strong link between the reweighting of individual companies and the sustainability data used to construct the index. Hence, this paper gives special attention to the strength of the relationship between the index weights and the variables used to define the constraints. Sustainability indices may also serve an important role in their capacity as investment benchmarks—an indirect impact mechanism noted by Kölbel et al. (2020).
